Editors' ChoiceAutism

Decoding Autism

See allHide authors and affiliations

Science Translational Medicine  18 Dec 2013:
Vol. 5, Issue 216, pp. 216ec208
DOI: 10.1126/scitranslmed.3008134

Autism spectrum disorders (ASDs) are a broad class of disorders that involve social deficits, language deficits, and restrictive/repetitive behaviors that range from severe to mild. A major role for genetics in ASDs is clear because heritability estimates are as high as 90%. However, we now understand that the majority of autistic phenotypes are influenced by hundreds of different genes with diverse functions that range from regulating synapse function to gene expression. Further, the variants involved are diverse and include additive effects from common variants, inherited copy number variants, as well as de novo mutations. Thus, despite a clear role for genetics, the complexity of ASDs has presented a major challenge. An enticing possibility is that the genes and variants that influence ASDs converge in some way to influence a common neural system, developmental process, and/or aspect of brain function. Two recent articles in Cell, one by Parikshak and colleagues and the other by Willsey and colleagues, have made major advances toward addressing this possibility.

Both studies have leveraged a growing repository of transcriptome data for different human brain regions and developmental stages. The authors constructed gene coexpression networks on the basis of correlated patterns of expression between genes. Parikshak et al. took an unbiased approach in which networks were constructed by using an established method that divides correlated and anticorrelated genes into coexpression modules. They discovered five coexpression modules that are enriched for ASD-linked genes and learned that ASD-linked modules exhibit enriched expression in the superficial layers (L2 to L4) of the cortex. Two ASD-linked modules were enriched for transcriptional regulators expressed at high levels during early development, and three modules were enriched for synaptic genes expressed at high levels postnatally. The Willsey et al. study took a different approach by defining correlated coexpression networks relative to nine high-confidence ASD-linked genes. Networks for these genes were constructed for four different brain region groupings across 13 developmental windows (52 spatiotemporal networks). The authors found four networks at specific developmental stages and in specific brain regions that were enriched for 122 probable ASD-linked genes. Two of the most highly enriched networks arose in the prefrontal cortex during midgestation, and these networks were enriched for genes in deep-layer (L5 or L6) projection neurons.

Parikshak et al.’s study predicts that alterations to upper-layer, glutamatergic cortical neurons are most likely to influence ASD-related phenotypes, whereas Willsey et al.’s study predicts that developmental disruptions to deep-layer glutamatergic neurons will have the greatest effects. Future functional imaging and model organism-based studies will help to test these predictions and advance our understanding of the genetic architecture of ASDs.

A. J. Willsey et al., Coexpression networks implicate human midfetal deep cortical projection neurons in the pathogenesis of autism. Cell 155, 997–1007 (2013). [Abstract]

N. N. Parikshak et al., Integrative functional genomic analyses implicate specific molecular pathways and circuits in autism. Cell 155, 1008–1021 (2013). [Abstract]

Stay Connected to Science Translational Medicine

Navigate This Article